U.S. patent application number 11/982492 was filed with the patent office on 2009-05-07 for wideband dichroic-filter design for led-phosphor beam-combining.
This patent application is currently assigned to Light Prescriptions Innovators, LLC. Invention is credited to Waqidi Falicoff.
Application Number | 20090116114 11/982492 |
Document ID | / |
Family ID | 40587837 |
Filed Date | 2009-05-07 |
United States Patent
Application |
20090116114 |
Kind Code |
A1 |
Falicoff; Waqidi |
May 7, 2009 |
Wideband dichroic-filter design for LED-phosphor beam-combining
Abstract
A general method is disclosed of designing two-component
dichroic short-pass filters operable for incidence angle
distributions over the 0-30.degree. range, and specific preferred
embodiments are listed. The method is based on computer
optimization algorithms for an N-layer design, specifically the
N-dimensional conjugate-gradient minimization of a merit function
based on difference from a target transmission spectrum, as well as
subsequent cycles of needle synthesis for increasing N. A key
feature of the method is the initial filter design, upon which the
algorithm proceeds to iterate successive design candidates with
smaller merit functions. This initial design, with high-index
material H and low-index L, is (0.75 H, 0.5 L, 0.75 H) m, denoting
m (20-30) repetitions of a three-layer motif, giving rise to a
filter with N=2m+1.
Inventors: |
Falicoff; Waqidi; (Stevenson
Ranch, CA) |
Correspondence
Address: |
DRINKER BIDDLE & REATH;ATTN: INTELLECTUAL PROPERTY GROUP
ONE LOGAN SQUARE, 18TH AND CHERRY STREETS
PHILADELPHIA
PA
19103-6996
US
|
Assignee: |
Light Prescriptions Innovators,
LLC
Altadena
CA
|
Family ID: |
40587837 |
Appl. No.: |
11/982492 |
Filed: |
November 2, 2007 |
Current U.S.
Class: |
359/588 ;
359/577; 359/586 |
Current CPC
Class: |
G02B 5/285 20130101 |
Class at
Publication: |
359/588 ;
359/586; 359/577 |
International
Class: |
G02B 5/28 20060101
G02B005/28 |
Goverment Interests
STATEMENT REGARDING FEDERALLY FUNDED RESEARCH AND DEVELOPMENT
[0002] This invention was supported in part by the National Energy
Technology Laboratory Award No. DE-FC26-05NT42341. The Government
may have certain rights in this invention.
Claims
1. A shortpass filter comprising alternating layers of two
materials, one of low index of refraction and the other of high
index of refraction, the filter having an effective index of
refraction of more than 95% of the index of refraction of the high
index material over the operating wavelength of said filter.
2. A thin-film dichroic filter according to claim 1 to receive
light over a range of small incidence angles from a medium, said
filter comprising alternating thin-film layers of high and low
refractive index on a substrate, the number and individual
thicknesses of said layers specified by a design, said design
determined by an optimization method comprising: providing a seed
filter with formula (0.75 H, 0.5 L, 0.75 H) m, said m being 20 to
40; defining a target-spectrum with a desired edge-wavelength and
at least one desired wavelength range, said edge-wavelength
specifying the transition from transmittance to reflectance, said
at least one desired wavelength range including a spectral
transmittance range below said edge-wavelength and a spectral
reflectance range above said edge-wavelength; optimizing the seed
filter by a method comprising repeated steps of generating and
evaluating a candidate filter design from the design of a previous
step, said steps starting with said seed filter and concluding with
said dichroic filter, said evaluating steps using a merit function
comprising a weighted sum over said at least one wavelength range
of the difference between said target-spectrum and the transmission
spectrum of said candidate filter design.
3. The dichroic filter of claim 2, wherein said optimization method
comprises successively applying a conjugate-gradient optimization,
a needle synthesis, a compaction, and a further conjugate-gradient
optimization.
4. The dichroic filter of claim 2, wherein said target spectrum
comprises a first transmission spectrum for a first incidence angle
within said range of small incidence angles and a second
transmission spectrum for a second incidence angle within said
range, the edge-wavelength of said second transmission spectrum
having a wavelength shift that is 40 to 60% that of the cosine
shift between said first incidence angle and said second incidence
angle.
5. The dichroic filter of claim 4, wherein said target spectrum
further comprises a third and a fourth transmission spectrum for a
respective third and fourth incidence angle, said four incidence
angles having weights summing to unity and representing the
expected angular distribution upon said filter, wherein said target
spectrum is the sum of the four said target spectra as combined by
said weights.
6. The dichroic filter of claim 2, further comprising a first layer
of said low refractive index receiving said light from said medium,
said first layer at least two wavelengths thick, said first layer
operable to prevent the transmission of light at high incidence
angles.
7. The dichroic filter of claim 2, wherein said weights are all
unit within said at least one desired wavelength range.
8. The dichroic filter of claim 7, wherein weights lower than said
unit weight are given to wavelengths in a range between said range
below said edge-wavelength and said range above said
edge-wavelength.
9. The dichroic filter of claim 2, wherein said weights are
proportional to the spectral distribution of a combination of blue
LED and yellow phosphor.
10. A dichroic filter comprising at least 78 layers having optical
thicknesses that are within 2% of the optical thickness of layers
in the same order in the filter of Table 1.
11. A dichroic filter operable upon unguided light transiting
therethrough, said filter comprising alternate layers of higher and
lower refractive index, wherein an outermost layer of said lower
refractive index has its thickness increased to at least the 2%
evanescent-wave coupling-decay depth.
12. A dichroic filter according to claim 11, comprising 80
consecutive layers, wherein at least 78 of the layers have optical
thicknesses that are within 2% of the optical thickness of the
corresponding layer in the filter of Table 2.
13. A dichroic filter according to claim 11, comprising consecutive
layers, wherein at least 97% of the layers have optical thicknesses
that are within 2% of the optical thickness of the corresponding
layer in the filter of Table 3.
14. A dichroic filter according to claim 11, comprising consecutive
layers having optical thicknesses that are within 2% of the optical
thickness of the corresponding layer in the filter of Table 4.
15. A non-polarizing short-pass dichroic filter consisting
essentially of alternating layers of Ta.sub.2O.sub.5 and of a
material having a refractive index not greater than SiO.sub.2, the
filter having an effective index of refraction greater than
1.95.
16. A non-polarizing short-pass dichroic filter according to claim
2, wherein there are at least three layers of Ta.sub.2O.sub.5
thicker than 300 nm.
17. A non-polarizing short-pass dichroic filter comprising of an
alternating stack of two materials, one of low index of refraction
and the other of high index of refraction; wherein at least 3
percent of the high index layers of said filter have a thickness
not less than 300 nm and the filter has an effective index of
refraction greater than 1.95.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application is filed concurrently with U.S. patent
application Ser. No. ______ (docket No. 047654-0018-00-US) of
Falicoff et al. for "Optical Manifold," the disclosure of which is
incorporated herein by reference in its entirety.
BACKGROUND OF THE INVENTION
[0003] The present invention relates generally to multilayer
interference filters with thin-film specifications designed by a
computer optimization program. The design methods disclosed herein
related specifically to the commercial design package called the
Essential Macleod, by the Thin Film Center, Inc. of Tucson Ariz.,
but could as well be performed by similar software packages or by a
customized optimizer program.
[0004] In the field of fiber optics, the use of a multiplicity of
discrete wavelengths arises in Wavelength Division Multiplexing
(WDM), but those filters only need to operate over narrow ranges of
wavelength and incidence angle. In illumination optics, however,
the range of wavelengths and angles is much wider, making an
efficient design much more difficult to generate. In general,
optimization programs need to begin not far from a viable solution,
so that proper selection of the initial trial design is vital to
design success. Also, optimizers are easily stalled by excessive
constraints, which can easily happen when the filter must operate
at a wide range of incidence angles, as is typical in an
illumination system. This kind of filter design has not been
addressed by the prior art.
[0005] A new illumination task for a dichroic filter is combining
the blue light of an LED and the yellow light from a
photostimulated patch of phosphor into a single white output beam.
Because both the blue and yellow light as they emerge from the LED
and the phosphor are uncollimated, no presently-available filter
can efficiently combine their light until the range of incidence
angles has been restricted. The most geometrically convenient
beam-combination angle is 45.degree., but unless the light is
polarized only a narrow angular range can be accommodated without
loss of efficiency, typically 42-48.degree., whereas a realistic
illumination system will exhibit a 30-60.degree. range of angles
into the filter. This .+-.15.degree. range, however, is relatively
narrow as collimation range goes, and cannot be expected to be
smaller. While a 45.degree. angle of beam combination is well known
from RGB displays, it has not been possible to overcome the severe
polarization effects that occur at this angle. In order to overcome
this, several RGB display systems were developed where the
band-pass operates at less than 45.degree..
[0006] A well known system that operates at a nominal 30.degree. is
the so-called Philips color TV camera. This design became an
industry standard and is based on U.S. Pat. No. 3,202,039. Another
approach is given in U.S. Pat. No. 3,905,684. In this prior art the
band-pass filter operates at a nominal angle of 16.degree. and
therefore is an improvement over U.S. Pat. No. 3,202,039, as it is
much easier to design a short or long pass filter that operates
about this incidence angle. A new system for combining un-polarized
yellow and blue light based on a short-pass filter operating at a
nominal incidence angle of 15.degree. is revealed in the
above-referenced concurrently filed U.S. patent application. The
design method of the present invention was originally developed for
designing a short-pass filter for the system revealed in that
contemporaneously filed application. The method can, however, be
utilized wherever a wide-acceptance short-pass filter that operates
between 0 to 30.degree. is needed, such as in the system of U.S.
Pat. No. 3,905,684.
[0007] In order to achieve high efficiencies, the short-pass filter
employed in the contemporaneously filed application should, in
addition to its short-pass functionality, act as a high performance
reflector at incidence angles greater than 70.degree.. Such a
filter/reflector is not revealed in the prior art. It is well known
how to achieve a high reflectance using total internal reflectance
(TIR) by going from a high index of refraction layer to a low index
of refraction layer. However, in this approach the low index layer
must be thick enough to prevent frustrated total internal
reflection. In order to not see this effect, the low index layer
needs to be on the order of two wavelengths thick or for visible
light a dimension of over 1000 nm is needed. However, in thin film
designs the layers are typically a fraction of a wavelength thick,
and the outer layer of the stack typically has to be optimized to a
thickness to within a couple of nanometers to properly take into
account the combined interference effects of all the layers.
[0008] Surprisingly, it was determined that using the design method
described herein a solution is available in which at least one of
the two outer layers of the stack for the short-pass filter is a
low index layer, the thickness of which can be freely modified
without affecting the characteristics of the original short-pass
filter. By increasing the thickness of this low index layer
(typically Silicon Dioxide) to approximately 1200 nm, both
short-pass filter and reflector performance goals can be achieved.
TIR is achieved by having a layer of a higher index medium, such as
a glass cover or adhesive, before the thick low index first layer
on the side of the filter on which the light to be reflected is
incident, or in the other direction by having a high index layer
(layer number two in the stack) followed by a thick low index layer
(layer number one). The performance in the reflection mode is also
excellent in both directions. In one direction (where the light is
traveling from the substrate on which the filter is deposited,
through the stack (starting at the layer numbered 80 in the
Examples in the Tables below) to the thick layer (the layer
numbered 1 in Tables 2 and 3) the reflectance is theoretically
perfect at all angles above the critical angle for the materials.
In the other direction (where the light is traveling from the
adhesive or cover directly to the thick layer) the average
reflectance for the same incidence angles as the other direction in
the visible wavelengths is on the order of 99%. This is believed to
be a novel configuration and not taught in the design of thin films
in the prior art. In the configuration identified as optimal with
reference to FIG. 8 of the drawings of the concurrently filed
application, with the filter 8406 coated onto the underside of the
exit prism 8407, the direction of theoretically perfect reflectance
is the direction of the yellow light (500-700 nm), and the
direction of theoretically 99% reflectance is the direction of the
blue light (460 nm).
SUMMARY OF THE INVENTION
[0009] The afore-referenced concurrently filed patent application
discloses an LED-phosphor beam-combiner based on a 15.degree.
primary incidence angle, with a total angular range of
0-30.degree.. This is a relatively forgiving range of incidence
angles, because the wavelength shift of the filter's dichroic
performance curve of spectral reflectance (or transmittance) with
incidence angle is relatively small at low incidence angles. Both
the angular and polarization shifts are much smaller than for the
30-60.degree. range of the 45.degree. case. This reduced wavelength
shift of the 15.degree. embodiment contributes to ultra-high filter
efficiency (95%) because the minimum-energy overlap between the
blue LED spectrum and the yellow phosphor spectrum is at a
wavelength of approximately 500 nm, where there is a notch in the
resultant white spectrum.
[0010] Operating the filter at the notch wavelength (which can vary
slightly in accordance with specific LED and phosphor spectra) will
minimize the amount of inappropriately reflected long-wave blue
light and inappropriately transmitted short-wave yellow light. The
reflected blue light never reaches the phosphor, and the
transmitted yellow light never joins the exit beam, so that filter
efficiency is the product of both blue and yellow efficiencies,
requiring them both to be above 95%=97.5% to achieve 95% overall
efficiency, a strong challenge when considering the range of
incidence angles. As will be explained below, only the performances
at 15.degree. and 25.degree. are used in the merit function used in
optimizing the filter configuration. The inventor has found that
the performance at these two angles is an adequate proxy for the
overall performance within the range from 0.degree. to 30.degree.,
and that imposing more constraints tends to prevent a good
outcome.
[0011] Optical thin films are specified by a list of the layers,
beginning in the medium of the incident light, for which a design
goal has been established involving reflectance or transmittance as
a function of wavelength. The last layer in the list is the one
that is coated on the substrate, generally a substrate that is
transparent to all wavelengths of interest. Where the incident
light reaches the thin film filter through the substrate on which
the filter is coated, the stack could be computed with the
"adhesive" or "air" (depending on the design) as the "substrate" on
the back side and a solid "medium" on the incidence side. If the
resulting stack of layers is not reversible, the order of the
layers may be appropriately reversed in preparing the instructions
for coating onto the solid "medium." The reflectance of a
nonreversible stack can be different in one direction compared to
the other. A non-reversible stack may occur, for example, if the
thicknesses of the layers are not in a symmetrical pattern and one
or more of the materials used has an appreciable extinction
coefficient. Ta.sub.2O.sub.5 is an example of a material having an
extinction coefficient that is used in thin-film filters. This is a
preferred material in this invention because, unlike titanium
dioxide (which has a slightly higher index), it is highly
transparent in the blue wavelength and has lower internal stress.
These properties allow it to be used for designs that require many
layers. Tantalum Pentoxide is also very compatible with SiO2. This
combination has been used in the design and manufacture of filters
for WDM applications, where stack of hundreds of layers is not
uncommon.
[0012] Short-pass filters have the same transmission values in both
directions even if the stack is not symmetrical and one or more of
the materials has an extinction coefficient. The absorptance on the
other hand does change as a function of direction. An example of an
asymmetrical design having asymmetrical reflectance is provided in
FIG. 3.1 of the book "Optical Thin Films" by J. D. Rancourt, SPIE
Press 1996. Although the embodiments in this invention are
asymmetrical stacks and one of the materials has an extinction
coefficient, the major difference between the reflectance values in
the two directions for a given incidence angle is primarily a
consequence of the difference in the index of refraction of the
substrate and medium and the optical system coupled to the filter.
In most of the embodiments the substrate has a higher index of
refraction than the medium (in some they are the same). Therefore,
rays traveling toward the filter below the critical angle, will
emerge on the other side at a lower incidence angle. The change in
incidence angle can be calculated using Snell's law. In the
aforementioned contemporaneously filed application, the optical
system controls the angle of incidence of the rays striking the
filter from either direction, such that the difference is less than
the one based on Snell's law. This is why the reflectance can be
higher in the direction toward the filter from the side with higher
index material (substrate).
[0013] In deposition, the layers are successively deposited in the
reverse order of the list. When considering light going outwards
from the substrate, the reverse list can be used to analyze how it
will propagate therein. The preferred embodiments of the present
invention are two-component filters, comprising a low and a high
refractive index. The filter specification can be made as a
succession of letters, e.g., HLLLHL. Repetitions are designated as
(HL) m, for m repetitions of the same HL pair. The specification is
completed by listing the thicknesses of each layer, either in
physical units or optical thickness, relative to a reference
wavelength. Units of one quarter of the reference wavelength are
commonly used. The latter enables the same design to be scaled for
different wavelength bands and/or for materials of different
refractive index. For example, (0.75 H, 0.25 L) 10 would denote ten
pairs of layers, each consisting of a high index layer 0.1875
wavelengths thick and a low index layer 0.0625 wavelengths thick
where the wavelength is measured within the material in question,
perpendicular to the plane of the layer. The wavelength used is a
"reference wavelength," usually the wavelength at which an ideal
filter would switch from transmission to reflection. For a real
filter, the reference wavelength is substantially the midpoint of
the transition from transmission to reflection for the central ray
of the incident beam (in the embodiments, a 15.degree. ray).
[0014] The huge number of possible combinations of up to 100 layers
will highlight the value of a powerful optimization method when it
finds a highly meritorious design. Part of the power of any method
is its merit function and its initial, or seed, design. In general,
design insight and control do not come from using many different
refractive indices and thicknesses, but from using just the highest
and lowest, and with only one thickness, to make (HL) m. In
particular, highly efficient preferred embodiment short-pass
filters for the 15.degree. beam combiner are derived from an
optimization that began with the simple seed filter (0.75H, 0.5 L,
0.75H) m, where H is tantalum pentoxide (n=2.0462) and L is silicon
dioxide (n=1.46), and the repetition number m is preferably 20 to
30. As is explained in more detail below, this results in a stack
with the pattern HLHHLH-, etc., where HH is in effect a single
double-thickness (1.5.lamda./4) H layer or an optical thickness of
0.375.
[0015] The number of layers N in the stack can be calculated once m
is chosen, using the formula N=2 m+1. The filter for embodiments of
the abovementioned 15.degree. beam-combiner is short-pass, in that
blue light is transmitted and yellow light is reflected. This seed
filter, however, is band-stop, reflecting over a central range of
wavelengths and transmitting the shorter and longer wavelengths.
This seed filter exemplifies a first feature in the design methods
disclosed herein--using as a seed filter a known edge-filter design
with a spectrum that is the opposite of the one desired. This seed
design is revealed for example on Baumeister, Phillip W., "Optical
Coating Technology", SPIE Press, Birminghan Wash., 2004, pages
5-66, which is incorporated herein by reference in its entirety,
and is essentially a long-pass design with ripples on either side
of the transmission zone.
[0016] Also, Baumeister points out that having thicker high
refractive index layers decreases the angle shift of a thin film.
The design Baumeister provides, however, does not have a sharp
slope in the primary transition zone, an important feature of
embodiments of the short-pass filter described below. Nor does
Baumeister provide any guidance as to how to make the long-pass
filter achieve a tight slope while maintaining a low angle shift.
Finally, the design in Baumeister is in fact for a filter which
operates in air, yet the angle shift he shows is on par with the
designs revealed herein, which operate inside a dielectric and
where the media and substrate both have a high index of refraction.
It is well known to those skilled in the art of designing short or
long pass filters, that the angle shift can be reduced if the
filter operates in air. Assuming the substrate of Baumeister's
design has an index of refraction of 1.5, an incidence angle of
35.degree. in air would be equivalent in terms of performance to an
incidence angle of 22.degree. inside the dielectric, reducing the
cosine shift. With a filter in air there is refraction at the
air/filter interface, so that the actual direction of the rays
through the filter is steeper than the external angle of incidence.
However, as is explained below, the "corrected" cosine shift for
the design in Baumeister remains more than the cosine shift
achievable by the filters described in the present application.
[0017] Reference is also made to Macleod, H. Angus, `Thin Film
Optical Filters`, Institute of Physics Publishing, Philadelphia
Pa., 2001 and to U.S. Pat. No. 5,812,405 to Meredith, which are
incorporated herein by reference in their entirety.
[0018] A second feature is the use of a duplex target-spectrum,
consisting of two transmission spectra, for two incidence angles
within the desired range of small incidence angles, small here
meaning 30.degree. or less. The two angles used for the embodiment,
15.degree. and 25.degree., have an inherent cosine shift of 6%, or
30 nm for an edge-wavelength near 500 nm. The target spectra,
however, have only a 15 nm shift between them--500 nm at 15.degree.
and 485 at 25.degree., or half the cosine shift. This discrepancy
imposes an additional constraint as compared with the first target
spectrum alone, whereas specifying a full cosine shift would make
the second target spectrum redundant. By comparison the design from
Baumeister's book has an angle shift from 0.degree. to 22.degree.
(equivalent to 35.degree. in air) of 25 nm as can be seen in FIG.
5-88 in this book. This is more than double the 10 nm angle shift
of one embodiment described herein, as can deduced by comparing
curves 53 (0.degree.) and 57 (20.degree.) in FIG. 5a of this
specification. This clearly demonstrates the advantage of the new
design approach over prior art. Several of the new designs have
been manufactured using ordinary vacuum evaporation coating
equipment and the performance of the samples replicate the
theoretical values. The overall performance of these samples was
found to be on par with the finest rugate filters, or for some
performance categories, outperforming them. The equipment needed to
produce a rugate filter is very sophisticated as it must
continuously vary the index of refraction of the layers. Such
filters are therefore very expensive to manufacture and can be
produced by only a handful of companies in the world. By
comparison, the designs revealed herein are easily produced with
industry-standard coating equipment.
[0019] Another parameter used to predict and characterize the angle
shift of thin filters is the so-called effective index of
refraction n.sub.eff. The higher is its value, the lower is the
angle shift for a given angle of incidence as compared with the
value at zero incidence angle. The equation gives the new position
for the start of a short-pass or long-pass transition zone
.lamda..sub..theta. as a function of the transition position on the
spectral response curve at zero incidence angle .lamda..sub.0, the
angle of incidence of a ray .theta., and the effective index of
refraction n.sub.eff. The equation is:
.lamda. .theta. = .lamda. 0 1 - ( sin ( .theta. ) n eff ) 2
##EQU00001##
[0020] The transition feature most commonly quoted in the industry
and used in the above equation is the 50% transmission point. This
equation is given by several industry sources such as the 2007-2008
Catalog of Semrock of Rochester, N.Y. A similar definition is
provided on the website of www.omegafilters.com. According to Dr.
Angus Macleod, who is a noted expert in the field of thin films,
the parameter n.sub.eff is implicitly defined by this equation
(private communication). There is no direct way of accurately
calculating this parameter independent of this equation. Dr.
Macleod said an approximate value can be estimated by taking the
square root of the product of the lowest and highest index
materials used in the stack. So for a stack with the materials
SiO.sub.2, n=1.46, and Ta.sub.2O.sub.5, n=2.05, the effective index
would be equal to 1.73. This is a typical value for a well designed
filter. Numbers above this value have superior performance.
[0021] Dr. Macleod also states that he believes that the value of
the effective index should not be higher than the highest index
material in the stack. So a non-polarizing design with an effective
index of refraction at or near the index of the highest material in
the stack is operating near the theoretical limit. In the case of a
stack using Ta.sub.2O.sub.5 with an index of refraction for most
wavelengths the theoretical limit would be approximately 2.14 (near
the limit for some depositions of Ta.sub.2O.sub.5).
[0022] The parameters provided by Semrock in their catalog are for
filters which operate in air. The designs in the present
specification are all immersed in a dielectric. Therefore, to use
this formula one must correct the angle of incidence in the
dielectric using Snell's law. For example an incidence angle of
39.97.degree. in air would be equivalent to an incidence angle of
25.degree. in a dielectric having an index of refraction 1.52
resulting in 15% less angle shift. Semrock claim to have the
highest performing short-pass and long-pass filters in the
industry, the so-called "Razor-Edge" filters. They give typical
values for the effective index of refraction for the Razor-Edge
long-pass filters as 2.08 for s-polarized light and 1.62 for
p-polarized light. For the case where the both polarization states
are included (the non-polarizing case) it can be assumed that the
lower of the two values would apply or at best the average of the
two or 1.85. They also state that these numbers, the tables and the
equations can be used for their other filters, such as the
Razor-edge short-pass filters. They state that "Because the designs
are so similar for all of the Razor-Edge filters design for normal
incidence (see page 38 of the 2007-2008 Semrock Catalog), the set
of curves in the graph can be applied approximately to any of the
filters." Additionally, they state "For LWP [long-pass filters]
filters, the edge associated with s- and p-polarized light shifts
more than the edge associated with s-polarized light, whereas for
short-wave-pass (SWP) filters the opposite is true." Therefore,
using an effective index of 1.85 for short-pass filter should be
optimistic value for Razor-Edge short-pass filters from Semrock.
These filters are recognized as being representative in the
industry as the "state-of-the-art".
[0023] Using the above equation for the embodiment in Table 1 of
this invention, we can estimate the effective index of the design
at the reference wavelength 520 nm, where the angle of incidence
shifts from 0.degree. to 25.degree. inside a dielectric with an
index of refraction of 1.52. In this instance the equivalent angle
of incidence is as stated previously, 39.37.degree.. The transition
zone for this non-polarized (includes both p and s states) design
moves from 514 nm at zero incidence to 486 nm at 25.degree. (inside
the dielectric). So in this case .lamda..sub.0 is equal to 514 nm,
.lamda..sub..theta. is 486 nm. Using 39.37.degree. for the
incidence angle and solving for the effective index of refraction
in the above equation, we obtain a value of 1.973. This value is
considerably better than the ones quoted by Semrock in their
catalog and is close to the theoretical limit for a stack using
SiO.sub.2 and Ta.sub.2O.sub.5 n=2.046. In this instance the index
of Ta.sub.2O.sub.5 is 2.046, an effective index of 1.973 is over
97% of the theoretical maximum value. Using a higher index for the
substrate of 1.64, as is the case in several of the designs in the
invention, the effective index would be even higher, 2.13.
(Slightly above the theoretical limit. Most likely, as a
consequence of the approximate nature of the equation and the
round-off to the nearest nm of the data input to the equation.) For
the design of Table 3, with a thick first layer of SiO.sub.2,
.lamda..sub.0 is equal to 513.5 nm, .lamda..sub..theta. is 485.5
nm, resulting in an effective index of 1.972 (inside a dielectric
of 1.52). The effective index of the design of Table 3 is the same
as the design of Table 1, 1.973. This illustrates the advantage of
the new design and design approach compared with the prior art. It
operates at or very near the theoretical limit in terms of
minimizing the angle shift for the given materials of SiO.sub.2 and
Ta.sub.2O.sub.5.
[0024] A further confirmation of this fact can be obtained if one
modifies the medium and substrate for the design of Table 1, where
air is now the medium, and the substrate has an index of 1.64. The
design still behaves as a high performance non-polarizing
short-pass filter with a few percent losses in the transmission
zone. The effective index is determined by the angle shift of the
filter from 0.degree. incidence to 39.37.degree., as the incident
rays are now in air. In this case .lamda..sub.0 is equal to 513 nm,
.lamda..sub..theta. is 486.5 nm, and the effective index is 1.99,
97.7% of the theoretical maximum value and well above the
short-pass filter designs of Semrock and the prior art.
[0025] If a higher index material than Ta.sub.2O.sub.5 is used in
the design, then the effective index could be increased, reducing
the angle shift. However, this may come at a price in performance
for other targets. Using TiO.sub.2, for example, the effective
index can be increased but at the expense of a reduced transmission
in the blue wavelengths compared to the designs using
Ta.sub.2O.sub.5. This may not a problem for some applications that
operate at the longer wavelengths.
[0026] Another way one can compare the performance of the
embodiments of this invention to the prior art, such as the best
short-pass filters from Semrock, is to directly compare the angle
shift of the transition zone. For example, on pg. 37 of the
aforementioned catalog, the top graph on the page shows the
relative wavelength shift as a function of the angle of incidence
where the relative wavelength is defined as the fraction,
.lamda./.lamda..sub..theta.. The higher this fraction (1.0
represents no angle shift), the lower angle shift for a given
incidence angle and therefore the higher performance. For the
design of Table 1, the relative wavelength shift is 486/514 or
approximately 0.95. This is for a corrected incidence angle of
39.37.degree.. This is the same value as the Semrock nonpolarizing
design when the incidence angle is 30.degree.. This is
approximately a 9.degree. improvement in performance over this
prior art.
[0027] The Semrock designs typically require many more layers than
the embodiments disclosed in the present specification. According
to the catalog, the Razor-Edge filters are covered by U.S. Pat. No.
7,068,430. An embodiment for a short-pass filter is provided in
Example 2 in this patent. The design consists of 180 layers
alternating between Ta.sub.2O.sub.5 and SiO.sub.2. This is 100
layers more than the designs from Tables 1 to 4 hereinbelow. In
addition, the manufacturing approach of U.S. Pat. No. 7,068,430
requires the technique of continuous optical monitoring, whereas
the embodiments in this invention can be manufactured using
standard crystal monitoring procedures, a much more common
approach. In fact other than the improved steepness of the edge of
the short-pass filter that is possible with using 180 layers, the
design of U.S. Pat. No. 7,068,430 does not outperform the designs
in this invention for a number of key parameters. For example, if
one compares the measured transmission curve 902 of FIG. 9 of U.S.
Pat. No. 7,068,430 with curve 53 of FIG. 5a of this invention
(representative of the actual measured values of this design), both
short-pass filters operating at an angle incidence of 0.degree.,
there is little difference in the two in the transmission zone. The
80 layer design may in fact have a slight edge over the other near
the transition zone. The improved steepness of the transition zone
of the 180 layer design is expected. The difference in the
reflection zone is also small as the 80-layer design achieves 99%
reflectance from 525 nm to nearly 700 nm.
[0028] It is well known to those who are skilled in the art of thin
film design that the width of the reflection zone can be widened by
increasing the number of paired layers in the filter. An 80-layer
stack is sufficient for handling approximately 200 nm, from 500 to
700 nm. (If more bandwidth is required then a design having more
than 80 layers can be chosen using the method described herein.)
Equations for calculating the bandwidth of the reflection zone of a
filter, as a function of the number of layer pairs, are given in
many texts on the subject. For example, pg. 63 of the book "Field
Guide to Optical Thin Films" by R. Willey, SPIE 2006 explains how
this is calculated. On pg. 64 of the same book it is shown how the
edge slope for bandpass filters "between the pass and block band of
an edge or bandpass filter (LWP or SWP)" is calculated. The figure
at the bottom of that page shows how the slope increases as the
number of pairs increase. (It only shows up to 20 pairs) So it
expected that a design of 180 layers can achieve a very steep slope
at the transition zone as well as having high reflectance over the
entire reflectance zone of the filter. The problem with the
approach used by U.S. Pat. No. 7,068,430 is that the angle shift
problem is not addressed. So the design works very well for
collimated light from a source such as a laser but it is not so
good when there is a wider beam angle such as 25.degree. to
30.degree. full angle. In this case the steep edge slope at each
angle of incidence does not help as the angle shift cancels out
much of this benefit. This issue is addressed in the present
invention.
[0029] The high-index material tantalum pentoxide used in the
preferred designs, although having a lower index than titanium
dioxide, has an advantage over this commonly used high index
coating material in that it has lower absorption in the blue
wavelengths. Therefore, it is preferred material when a design
calls for many thick layers. As was stated previously this is well
known to those who design filters for the WDM industry. To achieve
the highest performance for the preferred embodiments, the highest
purity tantalum pentoxide was used. In the applicants' tests, the
material used was PATINAL.RTM. tantalum pentoxide from EMD
Chemicals Inc. of Gibbstown, N.J.
[0030] The concurrently-filed patent application describes an
embodiment of a remote-phosphor white-light source that has a
5-32.degree. range of incidence angles upon the dichroic filter. In
addition, the filter totally internally reflects light on both
sides that is at grazing incidence angles (75.degree. or more). For
this added function, an embodiment is disclosed in the present
application with an added thick layer of low-index material, where
thick means that it suppresses the transmission of the evanescent
waves upon which the transmission of light by frustrated TIR
depends. Suppression of the evanescent wave sufficient that 98% or
99% of the light in the glancing rays is totally internally
reflected is achievable in certain embodiments.
[0031] According to one embodiment of the invention, there is
provided a thin-film dichroic short-pass filter comprising
alternating layers of two materials, one of low index of refraction
and the other of high index of refraction, and the filter has an
effective index of refraction of more than 95% of the index of
refraction of the high index material over the operating wavelength
of said filter.
[0032] According to another embodiment of the invention, the filter
may designed and manufactured by a method in which the number and
individual thicknesses of the layers are specified by a design. The
design is determined by an optimization method comprising providing
a seed filter with formula (0.75H, 0.5 L, 0.75H) m, m being 20 to
40. A short-pass target-spectrum is defined with a desired
edge-wavelength and at least one desired wavelength range, the
edge-wavelength specifying the transition from transmittance to
reflectance. The at least one desired wavelength range includes a
spectral transmittance range below the edge-wavelength and a
spectral reflectance range above the edge-wavelength, or vice versa
for a long-pass filter. The seed filter is optimized by a method
comprising repeated steps of generating and evaluating a candidate
filter design from the design of a previous step, the steps
starting with the seed filter and concluding with the dichroic
filter. The evaluating steps use a merit function comprising a
weighted sum over the at least one wavelength range of the
difference between the target-spectrum and the transmission
spectrum of the candidate filter design. The spectral transmittance
range and the spectral reflectance range may be separated by a
range including the edge wavelength that is ignored or given lower
weight.
[0033] According to a further embodiment of the invention, there is
provided a dichroic filter comprising at least 78 layers having
optical thicknesses that are within 2% of the optical thickness of
layers in the same order in the filter of any of Tables 1 to 4
below.
[0034] According to a further embodiment of the invention, there is
provided a dichroic filter operable upon unguided light transiting
therethrough, the filter comprising alternate layers of higher and
lower refractive index, wherein an outermost layer of said lower
refractive index has its thickness increased to at least the 2%
evanescent-wave coupling-decay depth.
[0035] According to a further embodiment of the invention, there is
provided a non-polarizing short-pass dichroic filter comprising an
alternating stack of two materials, one of low index of refraction
and the other of high index of refraction; wherein at least 3
percent of the high index layers of said filter have a thickness
not less than 300 nm and the filter has an effective index of
refraction greater than 1.95 and/or greater than 95% of the
refractive index of the high index of refraction material.
BRIEF DESCRIPTION OF THE DRAWINGS
[0036] The above and other aspects, features, and advantages of the
present invention will be more apparent from the following more
particular description thereof, presented in conjunction with the
following drawings wherein:
[0037] FIG. 1 shows the transmission spectrum of a 41-layer initial
design candidate, a band-stop reflector, at three incidence
angles.
[0038] FIG. 2 shows the spectrum at one angle along with its
variation as two layers are varied.
[0039] FIG. 3 shows the transmission spectrum of a 61-layer initial
design candidate, a band-stop reflector, at three incidence
angles.
[0040] FIG. 4 shows the spectrum at one angle of an interim design
after conjugate gradient optimization is carried on the 41-layer
initial design candidate of FIG. 1.
[0041] FIGS. 5A and 5B shows the spectrum of a preferred 80-layer
dichroic filter listed in Table 1.
[0042] FIG. 6A shows the effects of manufacturing variations in a
filter at 15.degree. incidence angle.
[0043] FIG. 6B shows the effects of manufacturing variations in a
filter at 25.degree. incidence angle.
[0044] FIG. 7 shows the effects of manufacturing variations in a
filter at 75.degree. incidence angle, for beam guidance.
[0045] FIG. 8 shows the effects of manufacturing variations in a
filter at 80.degree. incidence angle, nearly TIR.
[0046] FIG. 9 shows the transmission spectrum of a filter designed
with variable weights.
[0047] FIG. 10 shows a flow chart of a design method.
[0048] FIGS. 11A through 11D show graphs of four filters.
[0049] FIG. 12 shows a graph of nonuniform spectral weights.
DETAILED DESCRIPTION OF THE DRAWINGS
[0050] Reference will now be made in detail to various embodiments
of the present invention, examples of which are illustrated in the
accompanying drawings. The embodiments are described by way of
explanation, and not by way of limitation.
[0051] FIG. 1 shows a spectral transmittance graph 10 with
horizontal wavelength scale 11 in nanometers (nm) and vertical
transmittance scale 12 in %, for the m=20, 41-layer seed filter
(0.75H, 0.5 L, 0.75H) 20. Solid curve 13 is the transmittance for
15.degree., dashed curve 14 to its left is for 25.degree., and
dotted curve 15 to its right is for normal incidence. As may be
send in FIG. 1, there are minor changes in the shape of the curves,
showing that they do more than shift in frequency when the
incidence angle is changed.
[0052] FIG. 2 shows spectral graph 20 with horizontal wavelength
scale 21 in nanometers (nm) and vertical transmittance scale 22 in
%, for the m=20, 41-layer seed filter. Solid curve 23 is the
transmittance for 15.degree. (the same curve, for the same filter,
as curve 13 of FIG. 1). Dashed curve 24 is the transmittance for a
filter differing from the filter of curves 13 and 23 by a doubling
of the thickness of the first layer, and dotted curve 25 is the
transmittance for a filter differing from the filter of curves 13
and 23 by a doubling of the second layer. In the interests of
clarity, FIG. 2 shows, represented by the differences between the
three transmittance curves, the sensitivity curves for only the
first two of the N layers, rather than all of them, and at only one
frequency. Each of the N sensitivity curves represents a
wavelength-by-wavelength plot of the partial derivative of filter
response with respect to change in layer thickness. At each
wavelength these N sensitivity curves represent an N-dimensional
vector that is mathematically defined as the gradient, or direction
of maximum increase in filter transmittance.
[0053] The target filter value at that wavelength may require a
reduction in transmittance, so that the gradient is not always the
desired direction (which defines the `conjugate` gradient vector).
In conjugate-gradient filter optimization there is a set of weights
that expresses the relative importance of the various wavelengths
expected to be incident upon the filter. For example, weights could
be made proportional to the spectral emittance of a
photostimulative blue LED and of the phosphor it stimulates. At
each wavelength there is also the spectral transmittance of the
desired target filter. The conjugate-gradient technique utilizes
the known gradients across the wavelengths of filter operation,
weighted to combine all the gradients into one. From it can be
calculated the N changes in the layers (i.e. an N-vector) that will
maximize the change of the entire weighted filter curve towards the
target filter curve. The resultant set of layer changes is
implemented step by step to transform the initial seed filter into
the closest possible approximation of the target filter After that
it is time to add more layers, via the next step, of needle
synthesis. Needle synthesis is a software design technique by which
a thin layer (needle) of material is moved through a design and
inserted at the point where a merit function is most improved.
[0054] FIG. 3 shows spectral transmittance graph 30 with horizontal
wavelength scale 31 in nanometers (nm) and vertical transmittance
scale 32 in %, for the m=30, 61 layer seed filter using the seed
formula (0.75H, 0.5 L, 0.75H) 30 at the reference wavelength of 520
nm. Solid curve 33 is the transmittance for 15.degree. incidence
angle, dashed curve 34 to its left is for 25.degree., and dotted
curve 35 to its right is for normal incidence.
[0055] FIG. 4 shows spectral graph 40 with horizontal wavelength
scale 41 in nanometers (nm) and vertical transmittance scale 42 in
%, for the filter obtained after conjugate gradient optimization is
performed on the 41 layer seed design of m=30 of FIG. 3 to convert
the long-pass design to a short-pass using the described targets
for the conjugate gradient. One can also start the design process
using a value for m equal to 20 as shown in FIG. 1. The resulting
final designs will be nearly the same with minor variations in
thicknesses and the number of layers.
[0056] Solid curve 43 is the transmittance for 15.degree.. Note how
the shape of the filter has the characteristics of a short-pass
filter, albeit with some room for improvement. The targets used to
achieve this interim design are discussed in detail in the
description of the algorithm for FIG. 10.
[0057] FIG. 5A shows spectral graph 50 with horizontal wavelength
scale 51 in nanometers (nm) and vertical transmittance scale 52 in
%, for the final 80 layer dichroic filter shown in Table 1 after
needle synthesis and subsequent further conjugate gradient
optimization are performed. Solid curve 53 is the transmittance for
normal incidence, and barely differing from it is dashed curve 54
for 5.degree.. Close by is dotted curve 55 for 10.degree., broken
curve 56 for 15.degree., solid curve 57 for 20.degree., and
leftmost solid curve 58 for 25.degree.. This design assumes that
the medium and substrate surrounding the filter are the same
material with an index of 1.52. The design works nearly as well if
a higher index substrate is substituted such as PBM2Y glass from
Ohara Inc. of Kanagawa, Japan. If a new design is required with a
substrate having a higher index, the design of Table 1 can be
optimized using either conjugate gradient or simplex and replacing
the old substrate with the new. No change is needed to the
targets.
[0058] FIG. 5B shows curves 56 and 58 of FIG. 5a, the curves on
which the optimization is based, isolated for clarity. In FIG. 5b,
curve 56 is shown solid, and curve 58 to its left is shown dashed.
These curves ended up precisely where they were specified for the
optimization.
[0059] FIG. 6A shows spectral graph. 60 with horizontal wavelength
scale 61 in nanometers (nm) and vertical transmittance scale 62 in
%, for the final 80 layer dichroic filter of Table 1 at 15.degree.
incidence angle. Multiple curves 63 are for ten variations of the
design generated by randomly changing all layer thicknesses and
indices, with a Monte-Carlo normal distribution of sigma 1%. These
curves are representative of what would be expected of actual
samples manufactured at that error level.
[0060] FIG. 6B shows spectral graph 60 and multiple curves 64 for
25.degree. incidence.
[0061] FIG. 7 shows spectral graph 70 with horizontal wavelength
scale 71 in nanometers (nm) and vertical transmittance scale 72 in
% (but only to 40%), for the final 80 layer dichroic filter of
Table 3. Solid curve 73 is the transmittance for 75.degree. in the
direction of the glass medium, showing very small transmittance, as
desired for attaining high internal reflectance. In the substrate
direction with the higher index material (Ohara PBM2Y has an index
above 1.6) there is 100% reflectance for rays striking the filter
above 70.degree. incidence angle.
[0062] FIG. 8 shows spectral graph 80 with horizontal scale 81 and
vertical scale 82 (but only to 4%) for the same design. Solid curve
83 is the transmittance for rays at an 80.degree. incidence angle
traveling from the medium (index of 1.52) toward the filter. The
peak transmittance is a small spike of 3% at approximately 570 nm.
There is perfect TIR, to within the accuracy of the calculations,
in the reverse direction at all angles above 70.degree. incidence
angle.
[0063] FIG. 9 shows spectral graph 90 with horizontal wavelength
scale 91 in nanometers (nm) and vertical transmittance scale 92 in
%, for a dichroic filter designed with weights given by the spectra
of a typical blue LED and a typical yellow phosphor. Solid curve 93
is the transmittance for 15.degree. and dashed curve 94 to its left
is for 25.degree.. As compared with FIG. 5b, the cliffs are much
steeper, but at the price of some transmission around 700 nm, where
the human eye is insensitive, and some reflection around 400 nm,
where the blue LED has little light. Also, to the right of each
cliff are small spikes, allowed because there is little radiation
at these wavelengths. These artifacts can be removed by using unit
weights at all wavelengths between the LED's blue peak at 455 nm
and the phosphor's yellow peak at 540 nm.
[0064] FIG. 10 is a simplified flow chart of the algorithm. A more
detailed description follows. The algorithm 100 can be defined into
several distinct steps. First in step 101, the seed equation is put
into a suitable thin film design program. The inventors used the
computer software Essential Macleod from Thin Film Center of
Arizona. However any equivalent package can be used. In an
embodiment, the seed equation, in the notation explained above, is
(0.75 H, 0.5 L, 0.75 H) m, where m is greater than or equal to 20
and less than or equal to 40. The H in this formula represents one
quarterwave optical of tantalum pentoxide. Therefore, 0.75 H would
represent an optical thickness of 0.1875 wavelengths. This is
doubled for the interior layers of tantalum pentoxide to 0.375.
[0065] The L in the formula is one quarterwave optical. Therefore,
0.5 L is equal to an optical thickness of 0.125 wavelengths. In an
embodiment, the reference wavelength is set to 520 nm to replicate
the design of a short-pass filter for the inventors' remote
phosphor systems. For other applications a different reference
wavelength may be chosen. Next, in step 102 the targets are input
to the program. For a system which must accommodate incidence beam
angles from 0.degree. to 30.degree., one target may be set for
15.degree. incidence angle and the other at 25.degree.. (More
generally, the targets may be set at the mean of the two extreme
incidence angles and 5.degree. less than the highest incidence
angle.) For the 15.degree. target the transmission zone may extend
from 400 nm to 500 nm for a 520 nm reference wavelength. In the
embodiment, the target transmission for all these wavelengths is
set to 100%. The reflection zone may extend from 505 nm to 700 nm.
This short-pass filter design is for the visible range which
extends from 400 to 700 nm. However, the design method can be used
for other wavelengths including a device which operates in the UV
or infrared range. If this is required then the target for
transmission and reflection zones would be altered as required.
Also, as was stated earlier, the bandwidth of the reflection zone
can be widened by increasing the number of paired layers. So if the
design requires a wider bandwidth than is described above than the
number of layers will have to be increased above 80. In this case
the number of synthesis cycles in the needle optimization phase
needs to be increased
[0066] In the embodiment, the reflection target for all these
wavelengths is 100% for the two polarizations. The 25.degree.
transmission target zone may extend from 400 nm to 485 nm. This is
based on using half the cosine shift as described before. The
reflection zone may then start at 490 nm and extends to 700 nm. The
optimization in Essential Macleod and similar programs is based on
calculations at a finite number of wavelengths within the
reflection and transmission zones. The resolution of these points
may be either 1 nm or up to 2 nm and may be the same in all cases.
(For 1 nm setting this would mean there are one hundred target
points between 400 nm and 500 nm). The transmittance and
reflectance may be calculated as the mean of the p-polarization and
s-polarization in all cases. In step 102 the weights of all these
wavelengths can be left at the default value of one (all
wavelengths matter the same), or a custom set can be input to the
program.
[0067] FIG. 12 shows spectral graph 120 comprising horizontal
wavelength scale 121 in nm and vertical scale 122 running from zero
to one. Curve 123 shows the relative spectral distribution of a
blue LED, while curve 124 shows that of a yellow phosphor. In one
example, the weights are equal to curve 123 up to its peak, and
beyond the peak as far as the beginning of dashed curve 125. The
weights then follow dashed curve 125, which shows an example of
non-unity weights in between the LED and phosphor peaks. At the
long wavelength end, beyond the dashed curve 125, the weights would
follow curve 124.
[0068] Once the seed formula and targets are entered, optimization
is performed in step 103 using the known technique of the conjugate
gradient. For those not familiar with this technique there is an
excellent introduction to it in the article at the link
http://en.wikipedia.org/wiki/Conjugate_gradient_method, which is
incorporated herein by reference in its entirety. The optimization
settings for a program such as Essential Macleod for deriving the
interim design (as exemplified by the transmittance curve of FIG.
4) can be summarized as follows. The merit function power should be
set to 2. A power of two means the merit function is proportional
to the square of the weighted differences from target, or errors.
The limiting range for the merit function should be 0.0001 or
smaller. The maximum iterations should be set, to ensure that the
program terminates if it cannot find a solution with an acceptably
low merit value. The maximum number of iterations should be set to
at least 1000 but, not usually much higher than 1000. The
termination count should be set to 2. Once the interim design is
calculated, any zero thickness layers or those less than 4 nm or
0.01 of the Full Wave Optical Thickness (FWOT) should be removed
from the stack using the compacting feature of the software, shown
as step 104 in FIG. 10. Their minor contributions can be mostly
recovered by a subsequent conjugate gradient optimization, which
will adjust the remaining layers to make up for those removed by
compaction, although this is not shown.
[0069] Now the design is ready for needle synthesis, shown as step
105 in FIG. 10. The targets should remain the same as those
described earlier. The settings for needle synthesis optimization
are as follows. The merit power function, the limiting range for
the merit function, the maximum number of iterations, and the
termination match count should be set the same as for the conjugate
gradient optimization. The number of synthesis cycles should be set
high enough so that solutions are found for up to 100 layers (or
more as is required). A good starting number is 10 synthesis
cycles. The maximum number of new needles per pass should be set at
3, with a needle corresponding to one HLH layer from the original
seed. A program such as Essential Macleod will save a number of the
best designs. The designer may review them and decide which one
best fits the actual goal specifications, as opposed to which ones
have the lowest merit function. The merit function in the software
is not always the best indicator, because in the process shown in
FIG. 10 the target has been simplified by using 100% goals for each
wavelength.
[0070] An alternative target is described below, which takes into
account the relative importance of each wavelength and incidence
angle. However, it was also found that if one increases the number
and complexity of the targets quite often the software algorithms
cannot find a regional optimum. It is therefore beneficial to start
with the simple set of targets, arrive at a "high-quality" solution
based on these targets and then perhaps add more constraints to see
if there is a slightly better local optimum that improves on
it.
[0071] FIG. 9 illustrates how the "real" theoretical performance of
filter is assessed. A further example is provided in the
concurrently-filed application, where there is an actual merit
function for an optical system including the filter. Another way of
assessing the real merit of the filter is to plug the filter design
into a raytracing package such as Lightools or TracePro that can
model the combined effects of the filter, the optics and other
components of a system such as a phosphor.
[0072] It was found that the most practical solutions from a
manufacturing and performance standpoint comprise from 70 to 86
layers. Typically, the solutions with more than 80 layers have many
zero or near-zero thickness layers, and these can be compacted, in
step 106 of FIG. 10. The 80-layer solution shown in Table 1 was
derived from a compacted 86 layer solution found using needle
synthesis. Once these very thin or zero layers are removed, then
conjugate gradient optimization is used in step 107 of FIG. 10 to
arrive at a final design. The spectral evaluation of step 108
comprises the following. When searching for a potential best
solution one can search for the designs which have a number of
thick high index layers. For example, the design in Table 1 has
four layers of tantalum pentoxide greater than 300 nm thick. Those
are layer 4, 8, 20, 30. Also there are no thin layers in the
design. Designs having many thin layers can be more difficult to
manufacture. The thinnest layer in this design is approximately 22
nm but most are much thicker. Designs with a number of thin layers
(less than 10 to 15 nm) should not usually be considered.
[0073] Also, it is important to do an error calculation to see how
sensitive a design is to variation in the index and thicknesses for
the materials. Presently available manufacturing processes can
typically meet a specification of 1% standard deviation of the
thickness and index for both tantalum pentoxide and silicon
dioxide. Processes are available that can guarantee 1/2% standard
deviation for thin film coatings for these materials. FIGS. 6a and
6b illustrate the performance that should be attainable using
standard quality equipment and processes available from the thin
film industry. After step 108, a design is output at final step 110
of FIG. 10, or another round can be tried by step 109, changing the
targets or the target weights.
[0074] An important property of the 80-layer design of Table 1 (and
the family of filter designs derived from the methods described
herein) is that the first layer in the stack nearest the medium can
be varied in thickness without affecting its characteristics as a
short-pass filter. It was found that the first layer of silicon
dioxide can be freely modified from its thickness of approximately
94 nm to over 1000 nm without any significant change in
performance. This makes it possible to design the first low index
layer so that it can act as a TIR layer when coupled with the
dielectric medium with a higher index (such as an adhesive).
[0075] A TIR layer will reflect light that is beyond the critical
angle. For the case of SiO.sub.2 and the highest index
Ta.sub.2O.sub.5 this is sin.sup.-1 (1.46/2.14)=43.degree.. (In the
designs disclosed in the present specification, the Ta.sub.2O.sub.5
has a refractive index of 2.046 for the reference wavelength of 520
nm and the critical angle is approximately 46.degree.. The index of
refraction of Ta.sub.2O.sub.5 varies as a function of its purity as
well as other factors such as the deposition process and coating
equipment that is used. Also the value changes as a function of the
wavelength of light. Typically, each coater determines these values
based on a combination of information for the supplier or
manufacturer and the deposition process chosen for the production
of the filter. It is common practice for those skilled in art of
thin films that the wavelength variability of the index of the
materials must be included in the design process, as is the case
with the designs of Tables 1 to 4 below.) This property is useful
for several applications involving light that is guided within the
incident medium and traveling through the filter towards the
substrate. In the conventional design the thin first layer (and
other thin low index layers of the stack) allow evanescent coupling
across itself, defeating to a large extent TIR. Evanescent fields
decay rapidly with depth, so that any particular degree of decay,
say 1%, would correspond to a particular thickness for the
SiO.sub.2 first low-index layer. For guided light at incidence
angle .theta. in the medium (adjacent to the thick SiO.sub.2
layer), the evanescent coupling strength, through a layer of lower
index n.sub.1 and thickness h, between a layer of refractive index
no and one with the index n.sub.2 is given by (Baumeister, p.
2-44)
T=1/(1+F.sub.av)
F.sub.av=(F.sub.s+F.sub.p)/2
.beta.=2.pi.(h/.lamda.) [(n.sub.0 sin
.theta..sub.0).sup.2-n.sub.1.sup.2]
This is the average for the two polarizations:
F.sub.s=(1-A.sub.s).sup.2 cos h.sup.2.beta.+(B.sub.s+C.sub.s).sup.2
sin h.sup.2.beta.
A.sub.s=n.sub.2 cos .theta..sub.2/n.sub.0 cos .theta..sub.0
B.sub.s=n.sub.2 cos .theta..sub.2/n.sub.1.gamma.
.gamma.= [(n.sub.0 sin .theta..sub.0/n.sub.1).sup.2-1]
C.sub.s=n.sub.1.gamma./n.sub.0 cos .theta..sub.0
F.sub.p=(1-A.sub.p).sup.2 cos h.sup.2.beta.+(B.sub.p+C.sub.p).sup.2
sin h.sup.2.beta.
A.sub.p=n.sub.2sec.theta..sub.2/n.sub.0sec.theta..sub.0
B.sub.p=.gamma.n.sub.2sec.theta..sub.2/n.sub.1
C.sub.p=n.sub.1/.gamma.n.sub.0sec.theta..sub.0
[0076] For the case where light is guided within the medium
(n=1.52) at an incident angle of 75.degree. then n.sub.0=1.52,
.theta..sub.0=75.degree., n.sub.1=1.46, n.sub.2=2.046, and
.theta..sub.2=44.degree.. Solving for T=0.01 (where T is the
proportion of the incident light that is transmitted by the
evanescent wave) using the above equations from Baumeister, we
determine that h/.lamda.=1.59, which for blue light (.lamda.=460
nm) amounts to h=730 nm. This means that the first thick SiO.sub.2
layer must be 730 nm thick to reduce the transmission of a
75.degree. angle of incidence 460 nm blue light to 1% or less. Blue
LEDs typically have a maximum wavelength of approximately 515 nm.
So a more conservative value using Baumeister's equations for h is
820 nm.
[0077] The above calculation utilized Baumeister's set of formulae
for a three layer system, which does not take into account the
effect on transmittance of the remaining layers in the stack. In
reality a more complex model is required to accurately determine
the transmission. Using the Macleod software the entire stack of
the filter and not just the first three layers can be analyzed.
This software indicates that a thickness for the first layer of
approximately 1200 nm is needed to reduce the transmission to below
1% for the visible wavelengths. This is because in the full range
of wavelengths and angles, the filter's performance displays some
periodic behavior so that extra thickness will ensure low
transmittance for the entire range of angles and wavelengths. (A
discussion of this periodic behavior, which is called resonant
tunneling, is given in the book "Evanescent Waves from Newtonian
Optics to Atomic Optic" by F. de Formel, Springer 2001, pages
22-29.) The value of approximately 1200 nm is also obtained using
the simplified equation from Baumeister (two layers) for the case
when h/.lamda.>>1. In this instance the transmission is equal
to e.sup.(.about.2.beta.) (where .beta. is defined in the
aforementioned equation from Baumeister for a three-layer system),
which in this case for 1% transmission comes to a value of h of
1217 nm. Using this simplified two-layer Baumeister equation for
determining the thickness requirements for a 2% transmission, one
comes to a value for h of 1034 nm or nearly twice the wavelength
for the maximum wavelength of blue, 515 nm. The 2% value is the
preferred minimum transmission for incidence angles above the
critical angle for the short-pass filter embodiments of this
invention in the "blue" direction mentioned above (with the
incident beam approaching the thick layer of the stack through a
glue or cover layer, rather than through the remaining layers of
the stack). Ideally the transmittance in this direction should be
below 1% which for the case of the designs in Table 2 and Table 3,
this is achieved when the thickness of the first SiO2 layer is
greater than 1170 nm.
[0078] This design is shown in Table 2. Table 2 replicates the
design from Table 1 except for the change in the thickness of the
first layer in the stack (now 1200 nm) and a different material for
the substrate. Table 3 shows an optimized thickness for the first
layer after conjugate gradient is employed using the targets for
incidence angles less than 30.degree.. The value for the first
layer is just over 1172 nm, very close to the original value of
1200 nm.
[0079] The glass medium adjacent to layer 1 may be bonded to the
filter by an optically transparent adhesive, which may have a
refractive index of 1.52. By sandwiching the filter between two
materials each of which has a refractive index considerably greater
than that of the adjacent layer of the filter, TIR is provided for
light approaching from both sides. The primary purpose of the thick
first layer is to insure that evanescent waves cannot jump across
the filter. The thick layer 1 in Table 2 only improves the
reflectance marginally (up to 10% for some wavelengths) over the
design of Table 1. The design of Table 1 is quite good at
reflecting light at high incidence angles. In order for the thick
layer 1 of Table 2 to be effective, in the direction from the glass
medium through the adhesive towards layer 1, it is desirable for
the rays that enter the adhesive to propagate into the first layer
of the filter as incoherent light. Layer 1 then acts as a spacer.
The filter design may operate coherently even though the light from
the LEDs and the phosphor is incoherent. Except in the thick layer,
coherent operation does not detract from the operation of the
filter.
[0080] The design set out in Table 2 works very well. It was
determined, however, that a conjugate gradient optimization of this
design could improve its performance slightly, on the order of
0.5%. The results of this conjugate gradient optimization are shown
in Table 3. Note that the thickness of the first layer is now
reduced to just under 1200 nm. The design of Table 3 was arrived at
using the same target values as described earlier. Therefore, the
1200 nm guess was close to one of several local optimum solutions.
In this case the 1200 nm was chosen to begin with as it is thick
enough to eliminate losses arising from frustrated total internal
reflection. It was found that in the embodiments the thickness had
to be at least approximately twice the dominant wavelength of the
longest wavelength in the system to reduce transmission by
frustrated TIR to a reasonable level. For the phosphors used in the
devices of the concurrently filed application, the dominant
wavelength is between 550 and 570 nm, so a layer at least 1100 nm
thick is needed to reduce the evanescent wave transmission to
approximately 1% at high incidence angles (above 70.degree.).
[0081] Table 4 shows a further embodiment of a filter according to
the invention based on an 82 layer design. Except where specified,
the materials are SiO.sub.2 with a refractive index of 1.46132 and
an extinction index of less than 0.00001, and Ta.sub.2O.sub.5 with
a refractive index of 2.0462 and an extinction index of 0.00002 at
the reference wavelength of 520 nm. Both the substrate underlying
layer 82 and the medium overlying layer 1 are glass with a
refractive index of 1.52016 at the reference wavelength of 520 nm.
The geometric thickness is expressed in units of the reference
wavelength of light (520 nm) in air. The optical thickness is
expressed in units of the wavelength of the same light (520 nm) in
the medium of the specific layer.
[0082] FIGS. 11A through 11D show in graphical form the structures
of some of the filters disclosed herein. Graph 111 of FIG. 11A
shows the refractive index distribution of the various layers of
the seed filter of FIG. 1, their thicknesses expressed in terms of
the reference wavelength within the material. The dashed box shows
a single unit of the m=20 seed filter. Graph 112 of FIG. 1B shows
in the same format the structure for the intermediate filter of
FIG. 3 that was derived from the seed filter by the conjugate
synthesis method. Graph 113 of FIG. 11C shows in the same format
the filter of FIG. 5A. Graph 114 of FIG. 1D shows in the same
format the filter with the thick first layer 115, for the
aforementioned assurance of TIR of glancing-incidence guided
rays.
[0083] Although specific embodiments have been described, various
modifications may be made without departing from the spirit and
scope of the invention as defined in the appended claims. For
example, specific examples of filters have been described for
selectively transmitting blue light from an LED and reflecting
yellow light from a phosphor, with a cut-off between transmittance
and reflectance at a reference wavelength of 520 nm. The examples
set forth in Tables 1 to 4 use SiO.sub.2 with a refractive index of
1.46132 and Ta.sub.2O.sub.5 with a refractive index of 2.0462 for
the reference wavelength of 520 nm. The combination of
Ta.sub.2O.sub.5 and SiO.sub.2 is presently preferred, because it
combines low mechanical stress in a stack with many layers and low
absorption in the blue wavelengths used in the embodiments.
[0084] However, materials with different refractive indices may be
used, and/or the cut-off may be set at a different reference
wavelength, by scaling the physical thickness of the layers so as
to keep constant the optical thicknesses of the layers. For example
SiO.sub.2 can be replaced with magnesium fluoride. Conjugate
gradient optimization can also be performed on this new design to
arrive at an even higher performing design. Typically, the number
of layers in a stack of a material such as magnesium fluoride is
limited as magnesium fluoride has a higher internal stress than
silicon dioxide. Some modern coating techniques can reduce this and
make it practical to use magnesium fluoride in an 80 or more layer
design.) The optical thicknesses given in Tables 1 to 4 are Full
Wave Optical Thickness (FWOT), where the FWOT of a layer is defined
as the index of refraction of the material (at the reference
wavelength) times the physical thickness divided by the reference
wavelength.
[0085] The previously described merit function comprised a weighted
sum of the squared deviations of a filter spectrum from the target,
at the primary incidence angle of 15.degree. upon the filter or, in
step 101 of the example of FIG. 10, at the incidence angles of
15.degree. and 25.degree.. Other numbers of angles may be used in
defining the merit function. For example, in the above-mentioned
concurrently filed patent application, FIGS. 14, 15, and 16
respectively graphed the angular distribution of three different
input light beams, each of which contributes differently to the
white output luminosity. The entire angular range of zero to
30.degree., if represented relative to flux, can satisfactorily be
reduced to evaluating the spectrum only at the four discrete angles
of 10.degree., 15.degree., 20.degree., and 25.degree., since 5 and
30.degree. each represent less than 1 percent of the total flux.
The percentage allocations to the four angles in this example are
12.5% for 10.degree., 33.2% for 15.degree., 30.5% for 20.degree.,
and 23.8% for 25.degree.. The merit function would then comprise
summing the merit functions at these angles, as weighted by these
percentages. In designing a filter for a different angular
distribution, the composite merit function can easily be changed to
accommodate the new filter goals. In each case the target spectrum
would only be shifted by half the expected cosine effect.
[0086] Although the embodiments are directed to a short-pass
filter, it is believed that a long-pass filter on the same
principles can be generated by changing the target goals in the
optimization process.
TABLE-US-00001 TABLE 1 SPECIFICATION OF 15.degree. DICHROIC FILTER
optimized for glass on both sides with index of 1.52, for a
reference wavelength of 520 nm. Optical Thickness Physical Layer
Material (FWOT) Thickness (nm) Medium Glass (index 1.52016) 1 SiO2
0.264164 94.00 2 Ta2O5 0.449233 114.16 3 SiO2 0.124656 44.36 4
Ta2O5 1.307223 332.2 5 SiO2 0.188907 67.22 6 Ta2O5 0.448459 113.97
7 SiO2 0.183887 65.43 8 Ta2O5 1.294322 328.93 9 SiO2 0.087735 31.22
10 Ta2O5 0.457809 116.34 11 SiO2 0.126039 44.85 12 Ta2O5 0.485605
123.41 13 SiO2 0.113348 40.33 14 Ta2O5 0.625342 158.92 15 SiO2
0.062843 22.36 16 Ta2O5 0.535336 136.04 17 SiO2 0.169741 60.4 18
Ta2O5 0.464952 118.16 19 SiO2 0.123351 43.89 20 Ta2O5 1.292052
328.35 21 SiO2 0.138418 49.25 22 Ta2O5 0.469835 119.4 23 SiO2
0.169507 60.32 24 Ta2O5 0.555836 141.25 25 SiO2 0.066416 23.63 26
Ta2O5 0.574226 145.93 27 SiO2 0.157035 55.88 28 Ta2O5 0.476249
121.03 29 SiO2 0.146218 52.03 30 Ta2O5 1.288547 327.46 31 SiO2
0.107126 38.12 32 Ta2O5 0.427757 108.71 33 SiO2 0.174046 61.93 34
Ta2O5 0.394724 100.31 35 SiO2 0.213582 76 36 Ta2O5 0.813509 206.74
37 SiO2 0.117413 41.78 38 Ta2O5 0.372484 94.66 39 SiO2 0.265164
94.36 40 Ta2O5 0.301024 76.5 41 SiO2 0.28646 101.93 42 Ta2O5
0.271136 68.9 43 SiO2 0.304811 108.46 44 Ta2O5 0.264185 67.14 45
SiO2 0.310724 110.57 46 Ta2O5 0.258614 65.72 47 SiO2 0.315512
112.27 48 Ta2O5 0.25972 66 49 SiO2 0.312079 111.05 50 Ta2O5
0.266624 67.76 51 SiO2 0.304774 108.45 52 Ta2O5 0.283904 72.15 53
SiO2 0.290975 103.54 54 Ta2O5 0.315063 80.07 55 SiO2 0.32961 117.29
56 Ta2O5 0.39388 100.1 57 SiO2 0.381229 135.66 58 Ta2O5 0.334847
85.09 59 SiO2 0.26389 93.9 60 Ta2O5 0.3568 90.67 61 SiO2 0.241874
86.07 62 Ta2O5 0.781276 198.55 63 SiO2 0.245975 87.53 64 Ta2O5
0.354218 90.02 65 SiO2 0.289481 103.01 66 Ta2O5 0.34612 87.96 67
SiO2 0.434923 154.76 68 Ta2O5 0.387296 98.42 69 SiO2 0.350876
124.86 70 Ta2O5 0.317394 80.66 71 SiO2 0.370086 131.69 72 Ta2O5
0.341444 86.77 73 SiO2 0.395878 140.87 74 Ta2O5 0.371005 94.28 75
SiO2 0.493059 175.45 76 Ta2O5 0.451129 114.65 77 SiO2 0.303067
107.84 78 Ta2O5 0.387119 98.38 79 SiO2 0.286841 102.07 80 Ta2O5
0.414657 105.38 Total thickness 29.70267 8516 Substrate glass
(index 1.52)
TABLE-US-00002 TABLE 2 SPECIFICATION OF 15.degree. DICHROIC FILTER
with thick first layer, for a reference wavelength of 520 nm.
Physical Layer Material Thickness (nm) Medium Glass (index 1.52016)
1 SiO2 1200 2 Ta2O5 114.16 3 SiO2 44.36 4 Ta2O5 332.2 5 SiO2 67.22
6 Ta2O5 113.97 7 SiO2 65.43 8 Ta2O5 328.93 9 SiO2 31.22 10 Ta2O5
116.34 11 SiO2 44.85 12 Ta2O5 123.41 13 SiO2 40.33 14 Ta2O5 158.92
15 SiO2 22.36 16 Ta2O5 136.04 17 SiO2 60.4 18 Ta2O5 118.16 19 SiO2
43.89 20 Ta2O5 328.35 21 SiO2 49.25 22 Ta2O5 119.4 23 SiO2 60.32 24
Ta2O5 141.25 25 SiO2 23.63 26 Ta2O5 145.93 27 SiO2 55.88 28 Ta2O5
121.03 29 SiO2 52.03 30 Ta2O5 327.46 31 SiO2 38.12 32 Ta2O5 108.71
33 SiO2 61.93 34 Ta2O5 100.31 35 SiO2 76 36 Ta2O5 206.74 37 SiO2
41.78 38 Ta2O5 94.66 39 SiO2 94.36 40 Ta2O5 76.5 41 SiO2 101.93 42
Ta2O5 68.9 43 SiO2 108.46 44 Ta2O5 67.14 45 SiO2 110.57 46 Ta2O5
65.72 47 SiO2 112.27 48 Ta2O5 66 49 SiO2 111.05 50 Ta2O5 67.76 51
SiO2 108.45 52 Ta2O5 72.15 53 SiO2 103.54 54 Ta2O5 80.07 55 SiO2
117.29 56 Ta2O5 100.1 57 SiO2 135.66 58 Ta2O5 85.09 59 SiO2 93.9 60
Ta2O5 90.67 61 SiO2 86.07 62 Ta2O5 198.55 63 SiO2 87.53 64 Ta2O5
90.02 65 SiO2 103.01 66 Ta2O5 87.96 67 SiO2 154.76 68 Ta2O5 98.42
69 SiO2 124.86 70 Ta2O5 80.66 71 SiO2 131.69 72 Ta2O5 86.77 73 SiO2
140.87 74 Ta2O5 94.28 75 SiO2 175.45 76 Ta2O5 114.65 77 SiO2 107.84
78 Ta2O5 98.38 79 SiO2 102.07 80 Ta2O5 105.38 Total thickness nm
9621.78 Substrate Ohara PBM2Y (index 1.64)
TABLE-US-00003 TABLE 3 SPECIFICATION OF 15.degree. DICHROIC FILTER
with thick first Layer and high index substrate - Optimized. For a
reference wavelength of 520 nm. Optical Physical Thickness
Thickness Layer Material (FW) (nm) Medium Glass (index 1.52016) 1
SiO2 3.294544 1172.34 2 Ta2O5 0.453105 115.15 3 SiO2 0.115554 41.12
4 Ta2O5 1.312515 333.55 5 SiO2 0.181668 64.65 6 Ta2O5 0.454386
115.47 7 SiO2 0.178276 63.44 8 Ta2O5 1.295143 329.13 9 SiO2
0.076595 27.26 10 Ta2O5 0.461416 117.26 11 SiO2 0.139693 49.71 12
Ta2O5 0.494367 125.63 13 SiO2 0.108853 38.73 14 Ta2O5 0.632527
160.74 15 SiO2 0.073275 26.07 16 Ta2O5 0.516255 131.2 17 SiO2
0.166048 59.09 18 Ta2O5 0.459759 116.84 19 SiO2 0.120485 42.87 20
Ta2O5 1.295123 329.13 21 SiO2 0.145272 51.69 22 Ta2O5 0.467058
118.69 23 SiO2 0.167105 59.46 24 Ta2O5 0.546826 138.96 25 SiO2
0.065526 23.32 26 Ta2O5 0.576811 146.58 27 SiO2 0.149928 53.35 28
Ta2O5 0.475386 120.81 29 SiO2 0.162004 57.65 30 Ta2O5 1.290078
327.85 31 SiO2 0.113489 40.38 32 Ta2O5 0.422965 107.49 33 SiO2
0.17384 61.86 34 Ta2O5 0.399308 101.48 35 SiO2 0.206718 73.56 36
Ta2O5 0.816412 207.47 37 SiO2 0.098089 34.9 38 Ta2O5 0.38185 97.04
39 SiO2 0.261172 92.94 40 Ta2O5 0.302322 76.83 41 SiO2 0.289851
103.14 42 Ta2O5 0.272644 69.29 43 SiO2 0.299372 106.53 44 Ta2O5
0.266796 67.8 45 SiO2 0.309719 110.21 46 Ta2O5 0.263113 66.86 47
SiO2 0.310277 110.41 48 Ta2O5 0.261801 66.53 49 SiO2 0.310382
110.45 50 Ta2O5 0.269747 68.55 51 SiO2 0.304496 108.35 52 Ta2O5
0.282865 71.88 53 SiO2 0.287176 102.19 54 Ta2O5 0.325145 82.63 55
SiO2 0.319789 113.79 56 Ta2O5 0.402244 102.22 57 SiO2 0.372109
132.41 58 Ta2O5 0.342856 87.13 59 SiO2 0.253781 90.31 60 Ta2O5
0.362982 92.24 61 SiO2 0.233227 82.99 62 Ta2O5 0.793522 201.66 63
SiO2 0.226974 80.77 64 Ta2O5 0.361124 91.77 65 SiO2 0.293101 104.3
66 Ta2O5 0.339067 86.17 67 SiO2 0.433278 154.18 68 Ta2O5 0.386122
98.13 69 SiO2 0.357264 127.13 70 Ta2O5 0.316764 80.5 71 SiO2
0.371986 132.37 72 Ta2O5 0.344307 87.5 73 SiO2 0.393522 140.03 74
Ta2O5 0.375552 95.44 75 SiO2 0.48701 173.3 76 Ta2O5 0.435389 110.65
77 SiO2 0.325682 115.89 78 Ta2O5 0.364108 92.53 79 SiO2 0.328316
116.83 80 Ta2O5 0.403615 102.57 Total thickness 32.72882 9589.31
Substrate Ohara PBM2Y (index 1.64)
TABLE-US-00004 TABLE 4 Physical Thickness Geometric Layer Material
(FWOT) (nm) Thickness Medium Glass (index 1.52016) 1 SiO2 0.271192
96.5 0.185579 2 Ta2O5 0.449383 114.2 0.219619 3 SiO2 0.123371 43.9
0.084424 4 Ta2O5 1.307427 332.26 0.638954 5 SiO2 0.188706 67.15
0.129133 6 Ta2O5 0.449303 114.18 0.219579 7 SiO2 0.181557 64.61
0.124241 8 Ta2O5 1.294377 328.94 0.632576 9 SiO2 0.086066 30.63
0.058896 10 Ta2O5 0.458692 116.57 0.224168 11 SiO2 0.128299 45.65
0.087797 12 Ta2O5 0.485485 123.38 0.237262 13 SiO2 0.112794 40.14
0.077186 14 Ta2O5 0.626103 159.11 0.305983 15 SiO2 0.062705 22.31
0.042909 16 Ta2O5 0.534231 135.76 0.261084 17 SiO2 0.170726 60.75
0.11683 18 Ta2O5 0.4649 118.14 0.227202 19 SiO2 0.123237 43.85
0.084332 20 Ta2O5 1.29219 328.38 0.631507 21 SiO2 0.138821 49.4
0.094997 22 Ta2O5 0.470282 119.51 0.229832 23 SiO2 0.168994 60.14
0.115644 24 Ta2O5 0.554281 140.86 0.270883 25 SiO2 0.066367 23.62
0.045416 26 Ta2O5 0.5751 146.15 0.281058 27 SiO2 0.156034 55.52
0.106776 28 Ta2O5 0.476187 121.01 0.232718 29 SiO2 0.148165 52.72
0.101391 30 Ta2O5 1.288473 327.44 0.629691 31 SiO2 0.107516 38.26
0.073575 32 Ta2O5 0.427536 108.65 0.208942 33 SiO2 0.173591 61.77
0.11879 34 Ta2O5 0.395505 100.51 0.193288 35 SiO2 0.211964 75.43
0.145049 36 Ta2O5 0.814157 206.9 0.397887 37 SiO2 0.116718 41.53
0.079871 38 Ta2O5 0.37371 94.97 0.182636 39 SiO2 0.26461 94.16
0.181075 40 Ta2O5 0.30106 76.51 0.147131 41 SiO2 0.285858 101.72
0.195615 42 Ta2O5 0.271104 68.9 0.132492 43 SiO2 0.304719 108.43
0.208522 44 Ta2O5 0.264757 67.28 0.12939 45 SiO2 0.309604 110.17
0.211865 46 Ta2O5 0.258972 65.81 0.126562 47 SiO2 0.31581 112.38
0.216112 48 Ta2O5 0.259765 66.01 0.12695 49 SiO2 0.312063 111.05
0.213548 50 Ta2O5 0.266691 67.77 0.130335 51 SiO2 0.304968 108.52
0.208693 52 Ta2O5 0.284064 72.19 0.138825 53 SiO2 0.290628 103.42
0.19888 54 Ta2O5 0.316569 80.45 0.154711 55 SiO2 0.327194 116.43
0.223902 56 Ta2O5 0.391724 99.55 0.19144 57 SiO2 0.38143 135.73
0.261017 58 Ta2O5 0.338078 85.92 0.165222 59 SiO2 0.264239 94.03
0.180821 60 Ta2O5 0.358376 91.07 0.175142 61 SiO2 0.242854 86.42
0.166188 62 Ta2O5 0.783385 199.08 0.382849 63 SiO2 0.244237 86.91
0.167134 64 Ta2O5 0.356479 90.59 0.174215 65 SiO2 0.28885 102.78
0.197663 66 Ta2O5 0.348669 88.61 0.170398 67 SiO2 0.432857 154.03
0.296209 68 Ta2O5 0.384848 97.8 0.188079 69 SiO2 0.352282 125.36
0.24107 70 Ta2O5 0.311959 79.28 0.152458 71 SiO2 0.363949 129.51
0.249054 72 Ta2O5 0.318639 80.98 0.155722 73 SiO2 0.392297 139.6
0.268453 74 Ta2O5 0.374965 95.29 0.183249 75 SiO2 0.484534 172.42
0.331572 76 Ta2O5 0.452891 115.09 0.221333 77 SiO2 0.342098 121.73
0.234101 78 Ta2O5 0.399403 101.5 0.195192 79 SiO2 0.2378 84.62
0.162729 80 Ta2O5 0.405055 102.94 0.197955 81 SiO2 0.271192 107.46
0.206655 82 Ta2O5 0.449383 105.52 0.202914 Substrate Glass (index
1.52016) Total Thickness 30.38167 8715.8 16.76115
* * * * *
References